I really think the golden age of consumer and prosumer access to LLMs is done. I have subs to Claude, ChatGPT, Gemini, and Perplexity. I am running the same chat (analyse and comment on a text conversation) with all 4 of them. 3 weeks ago, this was 100% Claude territory, and it was superb. Now it is lazy, makes mistakes, and just doesn’t really engage. This is absolutely measurable - responses used to be in-depth and pick up all kinds of things i missed, now i get half-hearted paragraphs, and active disengagement (“ok, it looks like you dont need anything from me”)
ChatGPT is absurd. It will only speak to me in lists and bullets, and will go over the top about everything (“what an incredible insight, you are crushing it!”).
Gemini is… the village idiot and is now 50% hallucinations.
Perplexity refuses to give me the kind of insights i look for.
I think we are done. I think that if you want quality, you pay enterprise prices. And it may be about compute, but it may also be about too much power for the peasants.
[link] [comments]